822402 - Fling friction is different for sync and async scrolling

Status: RESOLVED WORKSFORME

(Firefox OS Graveyard :: General, defect)

Description

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

1. Fling on the settings (sync)
2. Fling on the browser (async)

Expected result:
Settings fling should feel the same as browser fling

Actual result:
Browser fling feels heavy and slow

Comment 1

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

Attachment: Fling faster and show what you've got

Comment 2

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

Comment on attachment 694689 [details] [diff] [review]
Fling faster and show what you've got

The patch increases the fling speed but forces the scroll offset to overlap the painted region. This means that you can fling off the painted region but when it gets an update notification it forces back to the painted region. This dramatically reduces empty screen during fling problems.

Comment 3

[Ben Francis [:benfrancis]]

This is not a gaia patch.

Comment 4

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Comment on attachment 694689 [details] [diff] [review]
Fling faster and show what you've got

Works great!

Doug, do you have cycles to review?

Comment 5

[Doug Sherk (:drs) (inactive)]

Comment on attachment 694689 [details] [diff] [review]
Fling faster and show what you've got

Review of attachment 694689 [details] [diff] [review]:
-----------------------------------------------------------------

::: gfx/layers/ipc/AsyncPanZoomController.cpp
@@ +1082,5 @@
>    return requestAnimationFrame;
>  }
>  
> +gfx::Point CalculcateRectDistance(const gfx::Rect& aRect1, const gfx::Rect& aRect2) {
> +  gfx::Point point;

I'd prefer to call this distance.

@@ +1100,5 @@
> +void AsyncPanZoomController::ShowSomethingIfWeHaveNothing(const FrameMetrics& aViewportFrame)
> +{
> +  gfx::Rect visibleRect(mFrameMetrics.mScrollOffset,
> +                        gfx::Size(mFrameMetrics.mCompositionBounds.width,
> +                                  mFrameMetrics.mCompositionBounds.height));

mScrollOffset and mCompositionBounds are in different units.

@@ +1106,5 @@
> +    gfx::Rect(aViewportFrame.mDisplayPort.x * aViewportFrame.mDevPixelsPerCSSPixel,
> +              aViewportFrame.mDisplayPort.y * aViewportFrame.mDevPixelsPerCSSPixel,
> +              aViewportFrame.mDisplayPort.width * aViewportFrame.mDevPixelsPerCSSPixel,
> +              aViewportFrame.mDisplayPort.height * aViewportFrame.mDevPixelsPerCSSPixel) +
> +    aViewportFrame.mScrollOffset);

With the conversion to device pixels, mDisplayPort*mDevPixelsPerCSSPixel is in different units than mScrollOffset.

Comment 6

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

nrc has fixed the disparity.

Comment 7

[qichao]

(In reply to Anthony Jones (:kentuckyfriedtakahe) from comment #6)
> nrc has fixed the disparity.

Hello Jones:
Is there any patch that nrc could provide ?